Biogeosciences, 16, 1657–1671, 2019 https://doi.org/10.5194/bg-16-1657-2019 © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License.

Reviews and syntheses: Dams, water quality and tropical reservoir stratification

Robert Scott Winton1,2, Elisa Calamita1,2, and Bernhard Wehrli1,2 1Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, 8092 Zurich, Switzerland 2Eawag, Swiss Federal Institute of Aquatic Science and Technology, 6047 Kastanienbaum, Switzerland

Correspondence: Robert Scott Winton ([email protected])

Received: 13 December 2018 – Discussion started: 2 January 2019 Revised: 25 March 2019 – Accepted: 3 April 2019 – Published: 23 April 2019

Abstract. The impact of large dams is a popular topic in These changes and associated environmental impacts need environmental science, but the importance of altered water to be better understood by better baseline data and more so- quality as a driver of ecological impacts is often missing phisticated predictors of reservoir stratification behavior. Im- from such discussions. This is partly because information proved environmental impact assessments and dam designs on the relationship between dams and water quality is rel- have the potential to mitigate both existing and future poten- atively sparse and fragmentary, especially for low-latitude tial impacts. developing countries where dam building is now concen- trated. In this paper, we review and synthesize information on the effects of damming on water quality with a special focus on low latitudes. We find that two ultimate physical 1 Introduction processes drive most water quality changes: the trapping of sediments and nutrients, and thermal stratification in reser- As a global dam construction boom transforms the world’s voirs. Since stratification emerges as an important driver and low-latitude river systems (Zarfl et al., 2014) there is a se- there is ambiguity in the literature regarding the stratification rious concern about how competing demands for water, en- behavior of water bodies in the tropics, we synthesize data ergy and food resources will unfold. The challenge created and literature on the 54 largest low-latitude reservoirs to as- by dams is not merely that they can limit the availability of sess their mixing behavior using three classification schemes. water to downstream peoples and ecosystems, but also that Direct observations from literature as well as classifications the physical and chemical quality of any released water is of- based on climate and/or morphometry suggest that most, ten altered drastically (Friedl and Wüest, 2002; Kunz et al., if not all, low-latitude reservoirs will stratify on at least a 2011). Access to sufficient quality of water is a United Na- seasonal basis. This finding suggests that low-latitude dams tions Environment Programme sustainable development goal have the potential to discharge cooler, anoxic deep water, (UNEP, 2016), and yet the potential negative effects of dams which can degrade downstream ecosystems by altering ther- on water quality are rarely emphasized in overviews of im- mal regimes or causing hypoxic stress. Many of these reser- pacts of dams (Gibson et al., 2017). voirs are also capable of efficient trapping of sediments and Dams are often criticized by ecologists and biogeo- bed load, transforming or destroying downstream ecosys- chemists for fragmenting habitats (Anderson et al., 2018; tems, such as floodplains and deltas. Water quality impacts Winemiller et al., 2016), disrupting floodplain hydrologic cy- imposed by stratification and sediment trapping can be mit- cles (Kingsford and Thomas, 2004; Mumba and Thompson, igated through a variety of approaches, but implementation 2005; Power et al., 1996) and for emitting large amounts of often meets physical or financial constraints. The impending methane (Delsontro et al., 2011). Such impacts act against construction of thousands of planned low-latitude dams will the promise of carbon-neutral hydropower (Deemer et al., alter water quality throughout tropical and subtropical rivers. 2016). In contrast, scientists have committed less investiga- tive effort to documenting the potential impacts of dams on

Published by Copernicus Publications on behalf of the European Geosciences Union. 1658 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification water quality. In cases where investigators have synthesized knowledge of water quality impacts (Friedl and Wüest, 2002; Nilsson and Renofalt, 2008; Petts, 1986), the conclusions are inevitably biased towards mid/high latitudes where the bulk of case studies and mitigation efforts have occurred. While there is much to be learned from more thor- oughly studied high-latitude rivers, given the fundamental role played by climate in river and lake functioning, it is im- portant to consider how the low-latitude reservoirs may be- have differently. For example, the process of reservoir strat- ification, which plays a crucial role in driving downstream water quality impacts, is governed to a large degree by lo- cal climate. Additionally, tropical aquatic systems are more prone to suffer from oxygen depletion because of lower oxy- gen solubility and faster organic matter decomposition at higher temperatures (Lewis, 2000). Latitude also plays an important role in considering ecological or physiological re- sponses to altered water quality. Studies focused on cold- water fish species may have little applicability to warmer rivers in the subtropics and tropics. Low-latitude river systems are experiencing a high rate of new impacts from very large dam projects. A review of the construction history of very large reservoirs of at least 10 km3 below ±35◦ latitude reveals that few projects were launched between 1987 and 2000, but in the recent decade (2001–2011) low-latitude mega-reservoirs have appeared at a rate of one new project per year (Fig. 1). Given that ongo- ing and proposed major dam projects are concentrated at low Figure 1. Construction history of the world’s 54 largest reservoirs latitudes (Zarfl et al., 2014), a specific review of water quality located below ±35◦ latitude. Project year of completion data are impacts of dams and the extent to which they are understood from the International Commission on Large Dams (https://www. and manageable in tropical biomes is needed. icold-cigb.net/, last access: 1 November 2018). Project start data are Large dams exert impacts across many dimension; but approximate (±1 year) and based on either gray literature source in this review, we largely ignore the important, but well- or for some more recent dams, i.e., visual inspection of Google covered, impacts of altered hydrologic regimes. Instead, we Earth satellite imagery. Grand Ethiopian Renaissance abbreviated as GER. Volume in map legend is in cubic kilometers. focus on water quality, while acknowledging that flow and water quality issues are often inextricable. We also disregard the important issue of habitat fragmentation and the many acute impacts on ecosystems and local human populations arising from dam construction activities (i.e., displacement on physical and chemical processes within the reservoirs that and habitat loss due to inundation). These important topics may affect downstream water quality. have been recently reviewed elsewhere (e.g., Winemiller et Finally, we review off-the-shelf efforts to manage or mit- al., 2016; Anderson et al., 2018). igate undesired chemical and ecological effects of dams re- In order to understand the severity and ubiquity of wa- lated to water quality. The management of dam operations ter quality impacts associated with dams, it is necessary to to minimize downstream ecological impacts follows the con- understand the process of lake stratification, which occurs cept of environmental flows (eflows). The primary goal of because density gradients within lake water formed by so- eflows is to mimic natural hydrologic cycles for downstream lar heating of the water surface prevent efficient mixing. By ecosystems, which are otherwise impaired by conventional isolating deep reservoir water from surface oxygen, stratifi- dam-altered hydrology of diminished flood peaks and higher cation facilitates the development of low-oxygen conditions minimum flows. Although restoring hydrology is vitally im- and a suite of chemical changes that can be passed down- portant to ecological functioning, it does not necessarily stream. To address the outstanding question of whether low- solve water quality impacts, which often require different latitude reservoirs are likely to stratify and experience asso- types of management actions. Rather than duplicate the re- ciated chemical water quality changes, we devote a section cent eflow reconceptualizations (e.g., Tharme, 2003; Richter, of this study to predicting reservoir stratification. This analy- 2009; Olden and Naiman, 2010; O’Keeffe, 2018), we focus sis includes the largest low-latitude reservoirs and is focused our review on dam management efforts that specifically tar-

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1659

are not well understood. Collecting the reservoir depth pro- files necessary to generate this key information may be more difficult than simply analyzing water chemistry below dams. Dam tail waters with low oxygen or reduced compounds, such as hydrogen sulfide or dissolved methane, are likely to stem from discharged deep water of a stratified reservoir.

2.1.1 Changing thermal regimes

Even at low latitudes where seasonal differences are less than temperate climates, aquatic ecosystems experience wa- ter temperatures that fluctuate according to daily and annual Figure 2. Conceptual summary of the physical and chemical water thermal regimes (Olden and Naiman, 2010). Hypolimnetic quality effects of dams and how they affect aquatic ecology. releases of unseasonably cold water represent alterations to a natural regime. Although the difference between surface and deep waters in tropical lakes is typically much less than those get water quality, which includes both eflow and non-eflow at higher latitudes (Lewis, 1996), the differences are often actions. much greater than the relatively subtle temperature shifts of 3–5 ◦C that have been shown to cause serious impacts (King ◦ 2 Impacts of dams on river water quality et al., 1998; Preece and Jones, 2002). For example, at 17 south of the Equator, Lake Kariba seasonally reaches a 6 to ◦ The act of damming and impounding a river imposes a funda- 7 C difference between surface and deep waters (Magadza, mental physical change upon the river continuum. The river 2010). The ecological impacts of altered thermal regimes velocity slows as it approaches the dam wall and the created have been extensively documented across a range of river reservoir becomes a lacustrine system. The physical change systems. of damming leads to chemical changes within the reservoir, Many aquatic insects are highly sensitive to alterations in which alters the physical and chemical water quality, which thermal regime (Eady et al., 2013; Ward and Stanford, 1982), in turn leads to ecological impacts on downstream rivers and with specific temperature thresholds required for completion associated wetlands. The best-documented physical, chemi- of various life cycle phases (Vannote and Sweeney, 1980). cal and ecological effects of damming on water quality are Since macroinvertebrates form an important prey base for summarized in Fig. 2 and described in detail in this sec- fish and other larger organisms, there will be cascading ef- tion. In each subsection, we begin with a general overview fects when insect life cycles are disrupted. Fish have their and then specifically consider the available evidence for low- own set of thermal requirements, with species often fill- latitude systems. ing specific thermal niches (Coutant, 1987). Altered ther- mal regimes can shift species distributions and community 2.1 Stratification-related effects composition. Development schedules for both fish and in- sects respond to accumulated daily temperatures above or Stratification, i.e., the separation of reservoir waters into sta- below a threshold, as well as absolute temperatures (Olden ble layers of differing densities, has important consequences and Naiman, 2010). Fish and insects have both chronic and for river water downstream of dams. A key to understanding acute responses to extreme temperatures. A systemic meta- the impacts of dams on river water quality is a precise under- analysis of flow regulation on invertebrates and fish popula- standing of the depth of the reservoir thermocline/oxycline tions by Haxton and Findlay (2008) found that hypolimnetic relative to spillway or turbine intakes. At many high-head- releases tend to reduce abundance of aquatic species regard- storage hydropower dams, the turbine intakes are more than less of setting. 10–20 m deep to preserve generation capacity even under ex- There exist several case studies from relatively low lat- treme drought conditions. For example, Kariba Dam’s in- itudes suggesting that tropical and subtropical rivers are takes pull water from 20 to 25 m below the typical level of susceptible to dam-imposed thermal impacts. The Murray the reservoir surface, which roughly coincides with the typ- cod has been severely impacted by cold-water pollution ical depth of the thermocline. In more turbid tropical reser- from the Dartmouth Dam in Victoria, Australia (Todd et voirs, thermocline depth can be much shallower, for exam- al., 2005) and a variety of native fish species were simi- ple at Murum Reservoir in Indonesia where the epilimnion is larly impacted by the Keepit Dam in New South Wales, Aus- just 4 to 6 m thick. Unfortunately, the turbine intake depths tralia (Preece and Jones, 2002). In subtropical China, cold- are not typically reported in dam databases. Furthermore, for water dam releases have caused fish spawning to be de- many reservoirs, especially those in the tropics, the mixing layed by several weeks (Zhong and Power, 2015). In tropi- behavior and therefore the typical depth of the thermocline cal Brazil, Sato et al. (2005) tracked disruptions to fish re- www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1660 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification productive success 34 km downstream of the Três Marias stream of low-latitude dams will suffer from slower oxygen Dam. In tropical South Africa, researchers monitoring down- recovery. stream temperature-sensitive fish in regulated and unreg- In addition to the direct impact imposed by hypoxic reser- ulated rivers found that warm water flows promoted fish voir water when it is discharged downstream, anoxic bottom spawning, whereas flows of 3 to 5 ◦C cooler hypolimnetic waters will also trigger a suite of anaerobic redox processes water forced fish to emigrate (King et al., 1998). within reservoir sediments that exert additional alterations to water quality. Therefore, anoxia can also exert indirect chem- 2.1.2 Hypoxia ical changes and associated ecological impacts. Here we dis- cuss two particularly prevalent processes: phosphorus remo- Stratification tends to lead to the deoxygenation of deep bilization and the generation of soluble reduced compounds. reservoir water, because of heterotrophic consumption and a lack of resupply from oxic surface layers. When dam in- 2.1.3 Phosphorus remobilization and eutrophication takes are deeper than the oxycline, hypoxic water can be passed downstream where it is suspected to cause signifi- Phosphorus (P) is an important macronutrient. Its scarcity cant ecological harm. Oxygen concentrations below 3.5 to or limited bioavailability to primary producers often limits 5 mg L−1 typically trigger escape behavior in higher organ- productivity of aquatic systems. Conversely, the addition of isms, whereas only well-adapted organisms survive below dissolved P to aquatic ecosystems often stimulates eutroph- 2 mg L−1 (Spoor, 1990). A study of 19 dams in the southeast- ication, leading to blooms of algae, phytoplankton or float- ern United States found that 15 routinely released water with ing macrophytes on water surfaces (Carpenter et al., 1998; less than 5 mg L−1 of dissolved oxygen and 7 released water Smith, 2003). Typically, eutrophication will occur when P with less than 2 mg L−1 of dissolved oxygen (Higgins and is imported into a system from some external source, but in Brock, 1999). Hypoxic releases from these dams often lasted the case of lakes and reservoirs internal P loading from sed- for months and the hypoxic water was detectable in some iments can also be important. Most P in the aquatic envi- cases for dozens of kilometers downstream. Below the Hume ronment is bound to sediment particles where it is relatively Dam in Australia, researchers found that oxygen concentra- unavailable for uptake by biota, but anoxic bottom waters of tions reached an annual minimum of less than 50 % satura- lakes greatly accelerate internal P loading (Nurnberg, 1984). tion (well under 5 mg L−1), while other un-impacted refer- Iron oxide particles are strong absorbers of dissolved ;, but ence streams always had 100 % oxygen saturation (Walker under anoxic conditions, the iron serves as an electron ac- et al., 1978). Researchers recently observed similar hypoxic ceptor and is reduced to a soluble ferrous form. During iron conditions below the Bakun Dam in Malaysia with less reduction, iron-bound P also becomes soluble and is released than 5 mg L−1 recorded for more than 150 km downstream into solution where it can build up in hypolimnetic waters. (Wera et al., 2019). Although observations and experiments Water rich in P is then either discharged through turbines have demonstrated the powerful stress that hypoxia exerts or mixed with surface waters during periods of destratifica- on many fish species (Coble, 1982; Spoor, 1990), there exist tion. Therefore, sudden increases in bioavailable P can stim- few well-documented field studies of dam-induced hypoxia ulate algal and other aquatic plant growth in the reservoir disrupting downstream ecosystems. This is partly because epilimnion. Theoretically, discharging of P-rich deep water it can be difficult to distinguish the relative importance of could cause similar blooms in downstream river reaches, but dissolved oxygen and other correlated chemical and physi- we are not aware of any direct observations of this phe- cal parameters (Hill, 1968). Hypolimnetic dam releases con- nomenon. Typically, nutrient releases in low-latitude con- taining low oxygen will necessarily also be colder than sur- texts are thought of as beneficial to downstream ecosystems face waters and they may contain toxic levels of ammonia because they would counteract the oligotrophication imposed and hydrogen sulfide, so it was not clear which factor was by the dam through sediment trapping (Kunz et al., 2011). the main driver for the loss of benthic macroinvertebrate di- Although dams seem to typically lead to overall reduc- versity documented below a dam of the Guadalupe River in tions in downstream nutrient delivery (see “Oligotrophica- Texas (Young et al., 1976). tion”, Sect. 2.2.2), the phenomenon of within-reservoir eu- Regardless, regulators in the southern US found the threat trophication because of internal P loading has been exten- of hypoxia to be sufficiently serious to mandate that dam tail sively documented in lakes and reservoirs worldwide. In the waters maintain a minimum dissolved oxygen content of 4 absence of major anthropogenic nutrient inputs, the eutrophi- to 6 mg L−1 depending on temperature (Higgins and Brock, cation is typically ephemeral and is abated after several years 1999). These dams in the Tennessee Valley are on the north- following reservoir creation. A well-known tropical example ern fringe of the subtropics (∼ 35–36◦ N), but are relatively is Lake Kariba, the world’s largest reservoir by volume. For warm compared with other reservoirs of the United States. many years after flooding a 10 % to 15 % of the lake surface Since oxygen is less soluble in warmer water and gas trans- was covered by Kariba weed (Salvinia molesta), a floating fer is driven by the difference between equilibrium and actual macrophyte. Limnologists attributed these blooms to decom- concentrations, it follows that low-oxygen stretches down- posing organic matter and also gradual P release from inun-

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1661 dated soils exposed to an anoxic hypolimnion (Marshall and turbulence allows finer sediments to fall out of suspension. Junor, 1981). Blockage of sediments and coarse material drives two related Indeed, some characteristics of tropical lakes seem to impact pathways. The first is physical, stemming from the make them especially susceptible to P regeneration from the loss of river sediments and bedload that are critical to main- hypolimnion. The great depth to which mixing occurs (of- taining the structure of downstream ecosystems (Kondolf, ten 50 m or more) during destratification, a product of the 1997). The second is chemical; the loss of sediment-bound mild thermal density gradient between surface and deep wa- nutrients causes oligotrophication of downstream ecosys- ter, provides more opportunity to transport deep P back to tems including floodplains and deltas (Van Cappellen and the surface (Kilham and Kilham, 1990). This has led limnol- Maavara, 2016). ogists to conclude that deep tropical water bodies are more prone to eutrophication compared with their temperate coun- 2.2.1 Altered habitat terparts (Lewis, 2000). There is of course variability within tropical lakes. Those with larger catchment areas tend to The most proximate impact of sediment starvation is the en- receive more sediments and nutrients from their inflowing hancement of erosion downstream of dams from outflows rivers and are also more prone to eutrophication (Straskraba causing channel incision that can degrade within-channel et al., 1993). These findings together suggest that thermally habitats for macroinvertebrates and fish (Kondolf, 1997). stratified low-latitude reservoirs run a high risk of experi- Impacts also reach adjacent and distant ecosystems such encing problems of eutrophication because of internal P re- as floodplains and deltas, which almost universally depend mobilization. upon rivers to deliver sediments and nutrients to main- tain habitat quality and productivity. In addition to sedi- 2.1.4 Reduced compounds ment/nutrient trapping, dams also dampen seasonal hydro- logic peaks, reducing overbank flooding of downstream river Another ecological stressor imposed by hypoxic reservoir reaches. The combination of these two dam effects leads to water is a high concentration of reduced compounds, such as a major reduction in the delivery of nutrients to floodplains, hydrogen sulfide (H2S) and reduced iron, which limit the ca- which represents a fundamental disruption of the flood pulse, pacity of the downstream river to cope with pollutants. Suf- affecting the ecological functioning of floodplains (Junk et ficient dissolved oxygen is not only necessary for the sup- al., 1989). port of most forms of aquatic life, but it is also essential River deltas also rely on sediment delivered by floods and to maintaining oxidative self-purification processes within damming has led to widespread loss of delta habitats (Giosan rivers (Friedl and Wüest, 2002; Petts, 1986). Reduced com- et al., 2014). Sediment delivery to the Mekong Delta has al- pounds limit the oxidative capacity of river water by act- ready been halved and could drop to 4 % of baseline if all ing as a sink for free dissolved oxygen. The occurrence of planned dams for the catchment are constructed (Kondolf H2S has been documented in some cases in the tail waters et al., 2014a). Elsewhere in the tropics, dam construction of dams, but the co-occurrence of this stressor with low tem- has been associated with the loss of mangrove habitat, such peratures and hypoxia make it difficult to attribute the ex- as at the Volta estuary in Ghana (Rubin et al., 1999). The tent to which it causes direct ecological harm (Young et al., morphology of the lower Zambezi’s floodplains and delta 1976). Researchers investigating fish mortality below Greens were dramatically transformed by reduced sediment loads Ferry Dam in Arkansas, USA, found H2S concentrations of associated with the Cahora Bassa megadam in Mozambique 0.1 mg L−1 (Grizzle, 1981), well above the recognized lethal (Davies et al., 2001). With diminished sediment delivery and −1 concentrations of 0.013 to 0.045 mg L of H2S for fish enhanced erosion from rising sea levels, the future of many based on toxicological studies (Smith et al., 1976). Lethal coastal deltas is precarious, as most of the world’s medium concentrations of ammonia for fish are 0.75 to 3.4 mg L−1 and large deltas are not accumulating sediment fast enough of unionized NH3 (Thurston et al., 1983), though we are un- to stay above water over the coming century (Giosan et al., aware of specific cases where these thresholds have been sur- 2014). passed because of dams. At the very least, the presence of reduced compounds at elevated concentrations indicates that 2.2.2 Oligotrophication an aquatic system is experiencing severe stresses, which, if sustained, will be lethal to most macroscopic biota. Although the densely populated and industrialized water- sheds of the world typically suffer from eutrophication, dam- 2.2 Sediment trapping induced oligotrophication, through sediment and nutrient trapping, can also severely alter the ecological functioning of Dams are highly efficient at retaining sediments (Donald et rivers and their floodplains, deltas and coastal waters. Glob- al., 2015; Garnier et al., 2005; Kunz et al., 2011). As rivers ally, 12 % to 17 % of global river phosphorus load is trapped approach reservoirs, the flow velocity slows and loses the po- behind dams (Maavara et al., 2015); but in specific loca- tential to slide and bounce along sand and gravel, while lost tions, trapping efficiency can be greater than 90 %, such as www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1662 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification at Kariba Dam on the Zambezi River (Kunz et al., 2011) 2.3 Reversibility and propagation of impacts and the Aswan Dam on the Nile (Giosan et al., 2014). Such extreme losses of sediments and nutrients can cause serious One way to think of the scope of dam impacts on water qual- acute impacts to downstream ecosystems, though examples ity is in terms of how reversible perturbations to each variable are relatively scarce because predam baseline data are not may be. Sediment trapped by a dam may be irreversibly lost often available. from a river and even unregulated downstream tributaries are Most of the best-documented examples of impacts stem- unlikely to compensate. Temperature and oxygen impacts of ming from oligotrophication are from temperate catchments dams, in contrast, will happen gradually through exchange with important and carefully monitored fisheries. For exam- with the atmosphere as the river flows. The speed of recov- ple damming led to the collapse of a valuable salmon fishery ery will depend upon river depth, surface areas, turbulence in Kootenay Lake, British Columbia, Canada, through olig- and other factors that may provide input to predictive mod- otrophication (Ashley et al., 1997). The fishery was eventu- els (Langbein and Durum, 1967). Field data from subtropical ally restored through artificial nutrient additions. Oligotroph- Australia and tropical Malaysia suggest that, in practice, hy- ication may impose similar ecological impacts in tropical poxia can extend to dozens or hundreds of kilometers down- contexts, such as in southern Brazil where an increase in stream of dam walls (Walker et al., 1978; Wera et al., 2019). water clarity following the closure of the Eng. Sérgio Motta Where reaeration measures are incorporated into dam opera- Dam (Porto Primavera Dam) was associated with a shift in tions, hypoxia can be mitigated immediately or within a few fish communities (Granzotti et al., 2018). Impacts have been kilometers, as was the case in Tennessee, USA (Higgins and perhaps most dramatic in the eastern Mediterranean follow- Brock, 1999). A study in Colorado, USA, found that thermal ing the closure of the Aswan Dam. The Lake Nasser reser- effects could be detected for hundreds of kilometers down- voir, following its closure in 1969, began capturing the total- stream (Holden and Stalnaker, 1975). Regardless of the type ity of the Nile’s famously sediment-rich floodwaters, includ- of impact, it is clear that downstream tributaries play an im- ing some 130 million tons of sediment that had previously portant role in returning rivers to conditions that are more reached the Mediterranean Sea. In the subsequent years there “natural” by providing a source of sediment and flow of more was a 95 % drop in phytoplankton biomass and an 80 % drop appropriate water quality. Water quality impacts of dams are in fish landings (Halim, 1991). With dams driving rivers to- therefore likely to increase and become less reversible when ward oligotrophy, and land-use changes such as deforestation chains of dams are built along the same river channel or on and agricultural intensification, causing eutrophication, glob- multiple tributaries of a catchment network. ally most rivers face some significant change to trophic state.

2.2.3 Elemental ratios 3 How prevalent is stratification of low-latitude reservoirs? The attention to phosphorus and nitrogen can obscure the im- portance of other nutrients and their ratios. The element sili- Because the chemical changes in hypoxia and altered ther- con (Si), which is also efficiently sequestered within reser- mal regimes both stem from the physical process of reservoir voirs, is an essential nutrient for certain types of phyto- stratification, understanding a reservoir’s mixing behavior is plankton. The simultaneous eutrophication and damming of an important first step toward predicting the likelihood of wa- many watersheds has led to decreases in Si to nitrogen ra- ter quality impacts. Unfortunately, there exists ambiguity and tios, which tends to favor nonsiliceous species over diatoms misinformation in the literature about the mixing behavior of (Turner et al., 1998). In the river Danube efficient trap- low-latitude reservoirs. To resolve the potential confusion, ping of Si in reservoirs over several decades led to a shift we review literature on the stratification behavior of tropical in Black Sea phytoplankton communities (Humborg et al., water bodies and then conduct an analysis of stratification 1997), coinciding with a crash in an important and produc- behavior of the 54 most-voluminous low-latitude reservoirs. tive fishery (Tolmazin, 1985). Turner et al. (1998) document 3.1 Stratification in the tropics a similar phenomenon at lower latitude in the Mississippi Delta. Decreases in Si loading led to a drop in the abun- For at least one authority on tropical limnology, the funda- dance of copepods and diatoms relative to the total meso- mental stratification behavior of tropical lakes and reservoirs zooplankton population in the Gulf of Mexico (Turner et al., is clear. Lewis (2000) states that “Tropical lakes are fun- 1998). These community shifts may have important impli- damentally warm monomictic . . . ” with only the shallowest cations for coastal and estuarine fish communities and the failing to stratify at least seasonally, and that periods of de- emergence of potential harmful algal blooms. stratification are typically predictable events coinciding with cool, rainy and/or windy seasons. Yet, there exists confu- sion in the literature. For example, the World Commission on Large Dams’ technical report states that stratification in low- latitude reservoirs is “uncommon” (McCartney et al., 2000).

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1663

The authors provide no source supporting this statement, but The Hutchinson and Loffler (1956) classification is meant the conclusion likely stems from the 70-year-old landmark to be applied to “deep” lakes and therefore is not useful for lake classification system (Hutchinson and Loffler, 1956), discriminating between stratifying and nonstratifying reser- which, based on very limited field data from equatorial re- voirs based on depth. It does suggest that all sufficiently gions, gives the impression that tropical lakes are predomi- deep reservoirs (except those above 3500 m altitude) should nately either oligomictic (mixing irregularly) or polymictic stratify. Most of the reservoirs in our data set fall into an (mixing many times per year). The idea that low-latitude wa- “oligomictic” zone, indicating irregular mixing (Fig. 3) when ter bodies are fundamentally unpredictable or aseasonal, as available literature suggests that most would be better de- well as Hutchinson and Loffler’s (1956) approach of classi- scribed as monomictic, with a predictable season of deeper fying lakes without morphometric information critical to un- mixing. This finding reaffirms one of the long-running crit- derstanding lake stability (Boehrer and Schultze, 2008), has icisms of this classification scheme: its overemphasis on been criticized repeatedly over subsequent decades as addi- oligomixis (Lewis, 1983). tional tropical lake studies have been published (Lewis, 1983, We found that the Lewis (2000) classification system for 2000, 1973, 1996). And yet, the original misleading clas- tropical lakes correctly identified most of the reservoirs in our sification diagram continues to be faithfully reproduced in data set as monomictic; however, six relatively shallow reser- contemporary limnology text books (Bengtsson et al., 2012; voirs known to exhibit seasonal stratification were misclassi- Wetzel, 2001). Since much of the water quality challenges fied into polymictic categories (Fig. 4a). Five of these six associated with damming develop from the thermal and/or lie within a zone labeled “discontinuous polymictic,” which chemical stratification of reservoirs, we take a critical look refers to lakes that do not mix on a daily basis, but mix at the issue of whether low-latitude reservoirs are likely to deeply more often than once per year. The literature suggests stratify predictably for long periods. that these lakes would be better described as “monomic- tic.” We should note that Lewis’s (2000) goals in generating this diagram were to improve upon the Hutchinson and Lof- 3.2 The largest low-latitude reservoirs fler (1956) diagram for low-latitude regions and to develop a classification system that could be applied to shallow lakes. To assess the prevalence of prolonged reservoir stratification Lewis (2000) does not mathematically define the boundaries periods that could affect water quality, we reviewed and syn- of difference and describes them as “approximate” based on thesized information on the 54 most-voluminous low-latitude his expert knowledge, so it is not terribly surprising that there reservoirs. Through literature searches, we found descrip- appears to be some misclassifications. tions of mixing behavior for 32 of the 54. Authors described In a final stratification assessment, we compared the nearly all as “monomictic” (having a single well-mixed sea- known mixing behavior from literature to calculations of son, punctuated by a season of stratification). One of these densimetric Froude number (Fr) (see Supplement) for the 35 reservoirs was described as meromictic (having a deep layer reservoirs for which discharge and surface area data are avail- that does not typically intermix with surface waters) (Zhang able (Fig. 4b). We calculated Fr using two different values for et al., 2015). The review indicates that 30 reservoirs stratify depth: first, using mean depth by dividing reservoir volume regularly for seasons of several months and thus could ex- by area; and second using dam wall height as a proxy for perience the associated chemical and ecological water qual- maximum depth. While some authors have suggested that ei- ity issues, such as thermal alterations and hypoxia. The two ther value for depth can be used (Ledec and Quintero, 2003), exceptions are Brazilian reservoirs, Três Irmãos and Ilha our analysis suggests that this choice can have a strong im- Solteira, described by Padisak et al. (2000) to be “mostly pact on Fr calculations and interpretation. The ratio between polymitic” (mixing many times per year). On further investi- mean and maximum depth within our data set ranges from gation, this classification does not appear to be based on di- 0.1 to 0.4 with a mean of 0.23. This means that all Fr val- rect observations but is a rather general statement of regional ues could be recalculated to be roughly one-quarter of their reservoir mixing behavior (see Supplement). value based on mean depth. This is inconsequential for reser- We compared our binary stratification classification based voirs with small Fr, but for those with large Fr it can lead to on available literature to the results of applying reservoir data a shift across the classification thresholds of 0.3 and 1. For to three existing stratification classification schemes. First, example, two reservoirs in our data set exceeded the thresh- we consider the classic Hutchinson and Loffler (1956) clas- old of Fr = 1 to indicate nonstratifying behavior when us- sification diagram based on altitude and latitude. Second, ing mean depth, but they drop down into the weakly strati- we plot the data onto a revised classification diagram for fying category when maximum depth is used instead. Nine tropical lakes proposed by Lewis (2000) based on reservoir other reservoirs shift from weakly to strongly stratifying. So morphometry. Finally we apply the concept of densimetric which value for Fr better reflects reality? It is worth consid- Froude number which can be used to predict reservoir strati- ering that reservoirs are typically quite long and limnologists fication behavior (Parker et al., 1975) based on morphometry often break them down into subbasins, separating shallower and discharge. arms closer to river inflows from deeper zones close to the www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1664 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification

Figure 3. The 54 most-voluminous low-latitude reservoirs overlaid onto a lake classification diagram (redrawn from Hutchinson and Loffler, 1956). dam wall. Use of maximum depth for Fr calculations prob- logic patterns. On a seasonal scale, large dams often homoge- ably better reflects stratification behavior at the dam wall, nize the downstream discharge regime. An eflow approach to whereas average depth may better indicate behavior in shal- reservoir management could implement a simulated flood pe- lower subbasins that are less likely to stratify strongly. Since riod by releasing some reservoir storage waters during the ap- the deepest part of a reservoir is at the dam wall and because propriate season. Although the eflow approach has tradition- stratification in this zone is the most relevant to downstream ally focused on the mitigation of ecological problems stem- water quality, it is probably most appropriate to use maxi- ming from disrupted hydrologic regimes, there is a grow- mum depth in Fr calculations. The two reservoirs described ing realization that water quality (variables such as water as polymictic by Tundisi (1990) and Padisak et al. (2000) fall temperature, pollutants, nutrients, organic matter, sediments, into the intermediate category of weakly stratifying (when dissolved oxygen) must be incorporated into the framework using mean depth), but three others within this zone are re- (Olden and Naiman, 2010; Rolls et al., 2013). There is al- ported to exhibit strong stratification (Deus et al., 2013; Na- ready some evidence that eflows successfully improve water liato et al., 2009; Selge and Gunkel, 2013). Better candidates quality in practice. In the Tennessee valley, the incorpora- for nonstratifying members of this reservoir data set are Ya- tion of eflows into dam management improved downstream cyretá and Eng. Sérgio Motta; but unfortunately, we could dissolved oxygen (DO) and macroinvertebrate richness (Bed- find no description of their mixing behavior in the literature. naríkˇ et al., 2017). Eflows have also been celebrated for pre- A field study with depth profiles of these reservoirs could venting cyanobacteria blooms that had once plagued an es- dispel this ambiguity and determine whether all of the largest tuary in Portugal (Chícharo et al., 2006). These examples il- low-latitude reservoirs stratify on a seasonal basis. lustrate the potential for eflows to solve some water quality Overall, this exercise of calculating Fr for large low- impacts created by dams. latitude reservoirs seems to indicate that most, if not all, are Unfortunately, eflows alone will be insufficient in many likely to stratify. This is an important realization because it contexts. For one, flow regulation cannot address the issue points to a significant probability for downstream water qual- of sediment and nutrient trapping without some sort of cou- ity problems associated with deep-water releases. Further- pling to a sediment flushing strategy. Second, the issues of more, the bulk of evidence suggests that these reservoirs mix oxygen and thermal pollution often persist under eflow sce- during a predictable season and not irregularly throughout narios when there is reservoir stratification. Even if eflows the year or across years, indicating that under normal condi- effectively simulate the natural hydrologic regime, there is no tions these reservoirs should be able to stratify continuously reason why this should solve water quality problems as long for periods of at least a few months. as the water intake position is below the depth of the reservoir thermocline and residence time is not significantly changed. The solution to hypoxic, cold water is not simply more of it; 4 Managing water quality impacts of dams but rather engineers must modify intakes to draw a more de- sirable water source, or destratification must be achieved. To 4.1 Environmental flows address the problems of aeration and cold-water pollution, dam managers turn to outflow modification strategies or de- The most developed and implemented approach (or collec- stratification. tion of approaches; reviewed by Tharme, 2003) for the mit- igation of dam impacts is the environmental flow (eflow), which seeks to adjust dam releases to mimic natural hydro-

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1665

dissolved oxygen in the Tennessee Valley, United States, hy- dropower operators continuously monitor DO in large dam outflows. DO levels are managed by hydropower plant per- sonnel specializing in water quality, aeration and reservoir operations (Higgins and Brock, 1999). However, dams do not always function as designed, especially older construc- tions in regions with less regulation and oversight; and in such cases, retrofits or adjusted management strategies may be effective. For example, Kunz et al. (2013) suggest that hy- poxic releases from Zambia’s Itezhi-Tezhi Dam (built in the 1970s) could be mitigated by releasing a mixture of hypolim- netic and epilimnetic waters. This proposed action could help protect the valuable fisheries of the downstream Kafue Flats floodplain.

4.3 Thermal buffering

The problem of cold-water pollution, much like hypoxia, is Figure 4. Reservoir morphometry and stratification behavior. (a) Relationships between area and depth for 40 of the 54 world’s driven by reservoir stratification and thus can be addressed by most-voluminous reservoirs located below ±35◦ latitude. Data are similar management strategies (Olden and Naiman, 2010). from the International Commission on Large Dams (https://www. Most common multilevel intakes are designed so that out- icold-cigb.net/, last access: 1 August 2018) (we excluded 14 reser- flows are derived from an appropriate mixture of epi- and voirs because of missing surface area data). Stratification behavior hypolimnetic waters to meet a desirable downstream temper- classification is synthesized from literature: circle symbols indicate ature threshold (Price and Meyer, 1992). A remaining chal- that the reservoir has an extended, predictable season of stratifica- lenge is a lack of understanding of what the thermal require- tion and/or mixes deeply no more than once per year; triangle sym- ments are for a given river system. Rivers-Moore et al. (2013) bols refer to two Brazilian reservoirs that authorities suggest are propose a method for generating temperature thresholds for likely to be polymictic, but for which no direct observations ex- South African rivers based on time series data from dozens ist (see Tundisi, 1990; Padisak, 2000); for reservoirs indicated by of monitoring stations. Unfortunately, basic monitoring data cross symbols, no published information on stratification behavior appears in literature searches. Dashed lines and classification la- for many regions of the tropics is sparse and quite fragmen- bels are approximations proposed by Lewis (2000). (b) Reservoirs tary, which further complicates the establishment of ecologi- sorted by densimetric Froude number, which is a function of reser- cal thermal requirements. voir depth, length, volume and discharge (Parker et al., 1975). The vertical dashed lines at Fr = 1 and Fr = 0.3 indicate the expected 4.4 Sediment manipulation boundaries between strongly, weakly and nonstratifying reservoirs (Orlob, 1983). Small dots represent Froude numbers if maximum Sediment trapping by dams is not only a water quality prob- depth (height of dam wall) is used instead of mean depth as sug- lem, as we have discussed, but also a challenge for dam man- gested by Ledec and Quintero, 2003. Discharge data is from the agement because it causes a loss of reservoir capacity over Global Runoff Data Centre (https://www.bafg.de/GRDC, last ac- time. Thus, in order to maintain generation capacity, many cess: 1 August 2018); five dams were excluded because of missing managers of hydropower dams implement sediment strate- discharge data. gies, which include the flushing of sediments through spill- ways or sediment bypass systems. A recent review of sed- iment management practices at hydropower reservoirs pro- 4.2 Aeration vides a summary of techniques in use and evaluates their ad- vantages and limitations, including operations and cost con- Hypoxia of reservoir tail waters is a common problem im- siderations as well as ecological impacts (Hauer et al., 2018). posed by dams. As a result, various management methods Sediment management can be implemented at the catchment for controlling dissolved oxygen content in outflows exist, scale, within the reservoir itself or at the dam wall. Sediment ranging in cost-effectiveness depending on the characteris- bypass systems are regarded as the most comprehensive solu- tics of the dam in question (reviewed by Beutel and Horne, tion, but may be expensive or infeasible because of reservoir 1999). Options include turbine venting, turbine air injection, dimensions, or cause more ecological harm than they allevi- surface water pumps, oxygen injection and aerating weirs. ate (Graf et al., 2016; Sutherland et al., 2002). As the issue of dam-induced hypoxia has been recognized Typically, sediment flushing is practiced in an episodic for many decades, most modern dams incorporate some sort manner, creating a regime of sediment famine punctuated by of oxygenation design elements. Where rules strictly regulate intense gluts that are not so much a feast, but rather bury www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1666 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification downstream ecosystems alive (Kondolf et al., 2014b). Envi- large dams, and those that do not create a deep reservoir are ronmentally optimized sediment flushes show potential for not subject to stratification-related effects. Thus, it is tempt- minimizing risks where these are feasible, but are rare in ing to conclude that small-scale hydro will have minimal wa- practice. A limitation is that not all dams are designed to ter quality impacts, but without a systematic assessment it is allow for sediment flushes, or reservoir characteristics im- impossible to make a fair comparison with large-scale hy- posed by local geomorphology render them impractical. Fur- dropower (Premalatha et al., 2014). Our analysis is biased thermore, no amount of flushing is able to transport coarser towards large systems for the practical reason that larger sys- bedload material (i.e., gravel or larger) downstream. To com- tems are much more likely to be described in databases and pensate for lost bedload and sediments, some managers have studied by limnologists. Future studies on the environmental made the expensive effort of depositing loose gravel piles impacts of small hydropower systems would be valuable. onto river margins so that they can gradually be incorporated into the downstream sediment pool as sediment-starved wa- 5.3 Better predictions of reservoir stratification ter inevitably cuts into banks (Kondolf et al., 2014b). behavior An alternative to sediment management at the site of the dam is the restoration of sediment-starved floodplain or delta Our predictions of reservoir stratification behavior based wetlands, but this process is likely to be prohibitively expen- upon morphometric and hydrologic data, while helpful for sive in most if not all cases. Restoration of drowning Missis- understanding broad patterns of behavior, are not terribly sippi River Delta in Louisiana, United States, are estimated useful for understanding water quality impacts of a specific to cost USD 0.5 to 1.5 billion per year for 50 years (Giosan planned dam. It would be much more useful to be able to re- et al., 2014). liably predict the depth of the thermocline, which could be compared with the depth of water intakes to assess the likeli- hood of discharging hypolimnetic water downstream. Exist- 5 Further research needs ing modeling approaches to predicting mixing behavior fall into two categories: mathematically complex deterministic or 5.1 More data from low latitudes process-based models and simpler statistical or semiempiri- cal models. Deterministic models holistically simulate many It is telling that, in this review focused on low latitudes, we aspects of lake functioning, including the capability to pre- often had to cite case studies from the temperate zone. For dict changes in water quality driven by biogeochemical pro- example, we were able to locate one study describing eco- cesses. Researchers have used such tools to quantify im- logical impacts stemming from dam-driven oligotrophication pacts of reservoirs on downstream ecosystems (Kunz et al., at low latitudes (Granzotti et al., 2018). The simple fact is 2013; Weber et al., 2017), but they require a large amount that most of the tropics and subtropics lie far from the most of in situ observational data, which is often lacking for low- active research centers and there has been a corresponding latitude reservoirs. This data dependence also makes them gap in limnological investigations. Europe and the United unsuitable for simulating hypothetical reservoirs that are in a States have 1.5 to 4 measurement stations for water quality planning stage and thus they cannot inform dam environmen- per 10 000 km2 of river basin on average. Monitoring den- tal impact statements. A promising semiempirical approach sity is 100 times smaller in Africa (UNEP, 2016). Our review was recently published, proposing a ‘generalized scaling’ for found that of the 54 most-voluminous low-latitude reservoirs, predicting mixing depth based on lake length, water trans- 22 (41 %) have yet to be the subject of a basic limnologi- parency and Monin–Obukhov length, which is a function of cal study to classify their mixing behavior. Further efforts to radiation and wind (Kirillin and Shatwell, 2016). This model monitor river water quality and study aquatic ecology in reg- was tuned for and validated against a data set consisting of ulated low-latitude catchments are needed to elucidate the mostly temperate zone lakes, so it is unclear how well it can blind spots that this review has identified. be applied to low-latitude systems. If this or another semi- 5.2 Studies of small reservoirs empirical model can be refined to make predictions about the stratification behavior of hypothetical reservoirs being Compared with larger dams, the ecological impact of small planned, it could provide valuable information about poten- hydropower dam systems have been poorly documented. Al- tial risks of water quality impacts on ecosystems of future though small dams are likely to have smaller local impacts dams. than large dams, the scaling of impacts is not necessarily proportional. That is, social and environmental impacts re- lated to power generation may be greater for small dams than 6 Conclusions large reservoirs (Fencl et al., 2015). Generalizations about small hydroelectric systems are difficult because they come We have found that damming threatens the water quality of in so many different forms and designs. For example, nondi- river systems throughout the world’s lower latitudes, a fact version run-of-river systems will trap far less sediment than that is not always recognized in broader critiques of large

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1667 dam projects. Water quality impacts may propagate for hun- The data on reservoir size and morphometry are available in the dreds of kilometers downstream of dams and therefore may World Register of Dams database maintained by the International be a cryptic source of environmental degradation, destroying Commission on Large Dams, which can be accessed (for a fee) at ecosystem services provided to riparian communities. Unfor- https://www.icold-cigb.org/ (International Commission on Large tunately, a lack of predam data on low-latitude river chem- Dams, 2018). The data on discharge is available in the Global istry and ecology makes it a challenge to objectively quantify Runoff Data Centre, 56068 Koblenz, Germany, accessible at https://www.bafg.de/GRDC/ (Bundesantalt für Gewässerkunde, such impacts. Building the capacity of developing countries 2018). at low latitudes to monitor water quality of their river systems should be a priority. Seasonal stratification of low-latitude reservoirs is ubiqui- Supplement. The supplement related to this article is available tous and is expected to occur in essentially any large tropical online at: https://doi.org/10.5194/bg-16-1657-2019-supplement. reservoir. This highlights the risk for low-latitude reservoirs to discharge cooler and anoxic hypolimnetic waters to down- stream rivers depending on the depth of the thermocline rela- Author contributions. RSW, BW and EC developed the paper con- tive to turbine intakes. Of course, in the absence of a random- cept. RSW and EC extracted the data for analysis. BW provided ized sampling study it is difficult to assess whether the anec- mentoring and oversight. RSW produced the figures. RSW wrote dotes we have identified are outliers or part of a more general the original draft. All authors provided critical review and revisions. widespread pattern. Further research could investigate how common these problems are and assess the geographic or de- sign factors that contribute toward their occurrence. Competing interests. The authors declare that they have no conflict It is difficult to assess which of the water quality impacts of interest. are most damaging for two reasons. First, dams impose many impacts simultaneously and it is often difficult to disentangle which imposed water quality change is driving an ecologi- Acknowledgements. This work is supported by the Decision An- cal response, or whether multiple stressors are acting syn- alytic Framework to explore the water-energy-food Nexus in com- plex transboundary water resource systems of fast developing coun- ergistically. Second, to compare the relative importance of tries (DAFNE) project, which has received funding from the Eu- impacts requires a calculation of value which, as we have ropean Union’s Horizon 2020 research and innovation programme learned from the field of ecological economics (Costanza et under grant agreement no. 690268. The authors thank Luzia Fuchs al., 1997), will inevitably be controversial. It does appear that for providing graphical support for the creation of Fig. 3. Marie- water quality effects, which can render river reaches unin- Sophie Maier provided helpful feedback on figure aesthetics. habitable because of anoxia and contribute to loss of flood- Comments from two anonymous reviewers greatly improved the plain and delta wetlands through sediment trapping, exert a manuscript. greater environmental impact than dam disruption to connec- tivity, which only directly affects migratory species. The mitigation of water quality impacts imposed by dams Review statement. This paper was edited by Tom J. Battin and re- has been successful in places, but its implementation is de- viewed by two anonymous referees. pendent on environmental regulation and associated funding mechanisms, both of which are often limited in low-latitude settings. Environmental impact assessments and follow-up References monitoring should be required for all large dams. The fea- sibility of management actions depends upon the dam de- Anderson, E. P., Jenkins, C. N., Heilpern, S., Maldonado- sign and local geomorphology. Thus, solutions are typically ocampo, J. A., Carvajal-vallejos, F. M., Encalada, A. C., custom-tailored to the context of a specific dam. We expect and Rivadeneira, J. F.: Fragmentation of Andes-to-Amazon that as the dam boom progresses, simultaneous competing connectivity by hydropower dams, Sci. Adv., 4, 1–8, water uses will exacerbate the degradation of water quality https://doi.org/10.1126/sciadv.aao1642, 2018. in low-latitude river systems. Further limnological studies of Ashley, K., Thompson, L. C., Lasenby, D. C., Mceachern, L., data-poor regions combined with the development and vali- Smokorowski, K. E., and Sebastian, D.: Restoration of an inte- dations of water quality models will greatly increase our ca- rior lake ecosystem: The kootenay lake fertilization experiment, pacity to identify and mitigate this looming water resource Water Qual. Res. J. Can., 32, 295–323, 1997. Bednarík,ˇ A., Blaser, M., Matouš˚u, A., Hekera, P., and challenge. Rulík, M.: Effect of weir impoundments on methane dy- namics in a river, Sci. Total Environ., 584–585, 164–174, https://doi.org/10.1016/j.scitotenv.2017.01.163, 2017. Data availability. All data used to produce Figs. 1, 3 Bengtsson, L., Herschy, R. W., and Fairbridge, R. W.: Encyclope- and 4 are available in the ETH Zurich Research Collection dia of lakes and reservoirs: geography, geology, hydrology and (https://doi.org/10.3929/ethz-b-000310656; Winton et al., 2018). paleolimnology, Springer, 2012. www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1668 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification

Beutel, M. W. and Horne, A. J.: A review of the ef- Head Dams and Identify Patterns of Across-Dam Variation, 10, fects of hypolimnetic oxygenation on lake and reser- e0141210, https://doi.org/10.1371/journal.pone.0141210, 2015. voir water quality, Lake Reserv. Manage., 15, 285–297, Friedl, G. and Wüest, A.: Disrupting biogeochemical cycles – Con- https://doi.org/10.1080/07438149909354124, 1999. sequences of damming, Aquat. Sci., 64, 55–65, 2002. Boehrer, B. and Schultze, M.: Stratification of lakes, Rev. Geophys., Garnier, J., Némery, J., Billen, G., and Théry, S.: Nutrient dynamics 46, RG2005, https://doi.org/10.1029/2006RG000210, 2008. and control of eutrophication in the Marne River system: Mod- Bundesantalt für Gewässerkunde: Global Runoff Data Centre, avail- elling the role of exchangeable phosphorus, J. Hydrol., 304, 397– able at: https://www.bafg.de/GRDC/, last access: 1 August 2018. 412, https://doi.org/10.1016/j.jhydrol.2004.07.040, 2005. Carpenter, S. R., Caraco, N. F., Correll, D. L., Howarth, Gibson, L., Wilman, E. N., and Laurance, W. F.: How R. W., Sharpley, A. N., and Smith, V. H.: Nonpoint Green is “Green” Energy?, Trends Ecol. Evol., 32, 922–935, pollution of surface waters with phosphorus and nitro- https://doi.org/10.1016/j.tree.2017.09.007, 2017. gen, Ecol. Appl., 8, 559–568, https://doi.org/10.1890/1051- Giosan, L., Syvitski, J., Constantinescu, S., and Day, J.: Cli- 0761(1998)008[0559:NPOSWW]2.0.CO;2, 1998. mate change: Protect the world’s deltas, Nature, 516, 31–33, Chícharo, L., Chícharo, M. A., and Ben-Hamadou, R.: Use of a hy- https://doi.org/10.1038/516031a, 2014. drotechnical infrastructure (Alqueva Dam) to regulate planktonic Graf, W., Leitner, P., Hanetseder, I., Ittner, L. D., Dossi, F., and assemblages in the Guadiana estuary: Basis for sustainable water Hauer, C.: Ecological degradation of a meandering river by local and ecosystem services management, Estuar. Coast. Shelf Sci., channelization effects: a case study in an Austrian lowland river, 70, 3–18, https://doi.org/10.1016/j.ecss.2006.05.039, 2006. Hydrobiologia, 772, 145–160, https://doi.org/10.1007/s10750- Coble, D. W.: Fish Populations in Relation to Dis- 016-2653-6, 2016. solved Oxygen in the Wisconsin River, Trans. Am. Granzotti, R. V., Miranda, L. E., Agostinho, A. A., and Fish. Soc., 111, 612–623, https://doi.org/10.1577/1548- Gomes, L. C.: Downstream impacts of dams: shifts in ben- 8659(1982)111<612:FPIRTD>2.0.CO;2, 1982. thic invertivorous fish assemblages, Aquat. Sci., 80, 1–14, Costanza, R., d’Arge, R., de Groot, R., Farber, S., Grasso, M., Han- https://doi.org/10.1007/s00027-018-0579-y, 2018. non, B., Limburg, K., Naeem, S., O’Neill, R. V., Paruelo, J., Grizzle, J. M.: Effects of hypolimnetic discharge on Raskin, R. G., Sutton, P., and van den Belt, M.: The value of fish health below a reservoir, Trans. Am. Fish. the world’s ecosystem services and natural capital, Nature, 387, Soc., 110, 29–43, https://doi.org/10.1577/1548- 253–260, https://doi.org/10.1038/387253a0, 1997. 8659(1981)110<29:EOHDOF>2.0.CO;2, 1981. Coutant, C. C.: Thermal preference: when does an asset Halim, Y.: The impact of human alteration of the hydrological cycle become a liability?, Environ. Biol. Fish., 18, 161–172, on ocean margins, in Ocean margin processes in global change, https://doi.org/10.1007/BF00000356, 1987. edited by: Mantoura, R. F. C., Martin, J.-M., and Wollast, R., Davies, B. R., Beilfuss, R. D., and Thoms, M. C.: Cahora Bassa ret- 301–327, John Wiley & Sons, 1991. rospective, 1974–1997: effects of flow regulation on the lower Hauer, C., Wagner, B., Aigner, J., Holzapfel, P., Flödl, P., Lie- Zambezi River, Verhandlungen des Int. Verein Limnol., 27, dermann, M., Tritthart, M., Sindelar, C., Pulg, U., Klösch, 2149–2157, 2001. M., Haimann, M., Donnum, B. O., Stickler, M., and Deemer, B. R., Harrison, J. A., Li, S., Beaulieu, J. J., Delsontro, T., Habersack, H.: State of the art, shortcomings and future Barros, N., Bezerra-Neto, J. F., Powers, S. M., Dos Santos, M. A., challenges for a sustainable sediment management in hy- and Vonk, J. A.: Greenhouse Gas Emissions from Reservoir Wa- dropower: A review, Renew. Sust. Energ. Rev., 98, 40–55, ter Surfaces: A New Global Synthesis Manuscript, Bioscience, https://doi.org/10.1016/j.rser.2018.08.031, 2018. 66, 949–964, https://doi.org/10.1093/biosci/biw117, 2016. Haxton, T. J. and Findlay, C. S.: Meta-analysis of the impacts of Delsontro, T., Kunz, M. J., Wuest, A., Wehrli, B., and Senn, D. B.: water management on aquatic communities, Can. J. Fish. Aquat. Spatial Heteogeneity of Methane Ebullition in a Large Tropical Sci., 65, 437–447, https://doi.org/10.1139/f07-175, 2008. Reservoir, Environ. Sci. Technol., 45, 9866–9873, 2011. Higgins, J. M. and Brock, W. G.: Overview of Reser- Deus, R., Brito, D., Mateus, M., Kenov, I., Fornaro, A., Neves, voir Release Improvements at 20 TVA Dams, J. En- R., and Alves, C. N.: Impact evaluation of a piscicul- erg. Eng., 125, 1–17, https://doi.org/10.1061/(ASCE)0733- ture in the Tucuruí reservoir (Pará, Brazil) using a two- 9402(1999)125:1(1), 1999. dimensional water quality model, J. Hydrol., 487, 1–12, Hill, L. G.: Oxygen Preference in the Spring Cave- https://doi.org/10.1016/j.jhydrol.2013.01.022, 2013. fish, Chologaster agassizi Loren, Trans. Am. Fish. Donald, D. B., Parker, B. R., Davies, J. M., and Leavitt, P. R.: Nutri- Soc., 97, 448–454, https://doi.org/10.1577/1548- ent sequestration in the Lake Winnipeg watershed, J. Great Lakes 8659(1968)97[448:OPITSC]2.0.CO;2, 1968. Res., 41, 630–642, https://doi.org/10.1016/j.jglr.2015.03.007, Holden, P. B. and Stalnaker, C. B.: Distribution and abun- 2015. dance of mainstream fishes of the middle and up- Eady, B. R., Rivers-Moore, N. A., and Hill, T. R.: Rela- per Colorado River basins, 1967–1973, Trans. Am. tionship between water temperature predictability and Fish. Soc., 104, 217–231, https://doi.org/10.1577/1548- aquatic macroinvertebrate assemblages in two South 8659(1975)104<217:DAAOMF>2.0.CO;2, 1975. African streams, African J. Aquat. Sci., 38, 163–174, Humborg, C., Ittekkot, V., Cociasu, A., and Bodungen, B. https://doi.org/10.2989/16085914.2012.763110, 2013. V.: Effect of Danube River dam on Black Sea biogeo- Fencl, J. S., Mather, M. E., Costigan, K. H., and Daniels, M. chemistry and ecosystem structure, Nature, 386, 385–388, D.: How Big of an Effect Do Small Dams Have?? Using Ge- https://doi.org/10.1038/386385a0, 1997. omorphological Footprints to Quantify Spatial Impact of Low-

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1669

Hutchinson, G. E. and Loffler, H.: The thermal stratifi- Lewis, W. M.: A Revised Classification of Lakes Based cation of lakes, P. Natl. Acad. Sci. USA, 42, 84–86, on Mixing, Can. J. Fish. Aquat. Sci., 40, 1779–1787, https://doi.org/10.1073/pnas.42.2.84, 1956. https://doi.org/10.1139/f83-207, 1983. International Commission on Large Dams: World Register Lewis, W. M.: Basis for the Protection and Management of Dams, available at: https://www.icold-cigb.org/GB/world_ of Tropical Lakes, Lake Reserv. Manage., 5, 35–48, register/world_register_of_dams.asp, last access: 1 August 2018. https://doi.org/10.1046/j.1440-1770.2000.00091.x, 2000. Junk, W. J., Bayley, P. B., and Sparks, R. E.: The flood- Lewis, W. M. J.: Thermal Regime Implications for Tropical Lakes, pulse concept in river-floodplain systems, Proc. Int. Limnol. Oceanogr., 18, 200–217, 1973. Large River Symp. Can. Spec. Publ. Fish. Aquat. Sci., Lewis, W. M. J.: Tropical lakes?: how latitude makes a difference, 106, available at: http://www.royalcommission.vic.gov.au/ in: Perspectives in Tropical Limnology, edited by: Schiemer, F. finaldocuments/summary/PF/VBRC_Summary_PF.pdf (last and Boland, K. T., 43–64, SPB Academic Publishing, Amster- access: 18 April 2019), 1989. dam, 1996. Kilham, P. and Kilham, S. S.: Endless summer: internal load- Maavara, T., Parsons, C. T., Ridenour, C., Stojanovic, S., Dürr, H. ing processes dominate nutrient cycling in tropical lakes, H., Powley, H. R., and Van Cappellen, P.: Global phosphorus re- Freshwater Biol., 23, 379–389, https://doi.org/10.1111/j.1365- tention by river damming, P. Natl. Acad. Sci. USA, 112, 15603– 2427.1990.tb00280.x, 1990. 15608, https://doi.org/10.1073/pnas.1511797112, 2015. King, J., Cambray, J. A., and Impson, N. D.: Linked effects of dam- Magadza, C. H. D.: Environmental state of Lake Kariba and released floods and water temperature on spawning of the Clan- Zambezi River Valley: Lessons learned and not learned, Lake william yellowfish Barbus capensis, Hydrobiologia, 384, 245– Reserv. Manage., 15, 167–192, https://doi.org/10.1111/j.1440- 265, https://doi.org/10.1023/A:1003481524320, 1998. 1770.2010.00438.x, 2010. Kingsford, R. T. and Thomas, R. F.: Destruction of wetlands Marshall, B. E. and Junor, F. J. R.: The Decline of Salvinia- and waterbird populations by dams and irrigation on the Mur- Molesta on Lake Kariba Zimbabwe, Hydrobiologia, 83, 477– rumbidgee River in Arid Australia, Environ. Manage., 34, 383– 484, https://doi.org/10.1007/BF02187043, 1981. 396, https://doi.org/10.1007/s00267-004-0250-3, 2004. McCartney, M. P., Sullivan, C., and Acreman, M. C.: Ecosystem Kirillin, G. and Shatwell, T.: Generalized scaling of seasonal Impacts of Large Dams, International Union for Conservation of thermal stratification in lakes, Earth-Sci. Rev., 161, 179–190, Nature and Natural Resources and United Nations Environmen- https://doi.org/10.1016/j.earscirev.2016.08.008, 2016. tal Programme, 2000. Kondolf, G. M.: Hungry water: Effects of dams and gravel Mumba, M. and Thompson, J. R.: Hydrological and ecolog- mining on river channels, Environ. Manage., 21, 533–551, ical impacts of dams on the Kafue Flats floodplain sys- https://doi.org/10.1007/s002679900048, 1997. tem, southern Zambia, Phys. Chem. Earth, 30, 442–447, Kondolf, G. M., Rubin, Z. K., and Minear, J. T.: Dams on the https://doi.org/10.1016/j.pce.2005.06.009, 2005. Mekong: Cumulative sediment starvation, Water Resour. Res., Naliato, D. A. D. O., Nogueira, M. G., Perbiche-neves, G., and Un- 50, 5158–5169, https://doi.org/10.1002/2013WR014651, 2014a. esp, P.: Discharge pulses of hydroelectric dams and their effects Kondolf, G. M., Gao, Y., Annandale, G. W., Morris, G. L., in the downstream limnological conditions: a case study in a Jiang, E., Zhang, J., Cao, Y., Carling, P., Fu, K., Guo, Q., large tropical river (SE Brazil), Lake Reserv. Manage., 14, 301– Hotchkiss, R., Peteuil, C., Sumi, T., Wang, H.-W., Wang, 314, https://doi.org/10.1111/j.1440-1770.2009.00414.x, 2009. Z., Wei, Z., Wu, B., Wu, C., and Yang, C. T.: Sustain- Nilsson, C. and Renofalt, B. M.: Linking flow regime and water able sediment management in reservoirs and regulated rivers: quality in rivers: a challenge to adaptive catchment management, Experiences from five continents, Earths Futur., 2, 256–280, Ecol. Soc., 13, 18, 2008. https://doi.org/10.1002/2013EF000184, 2014b. Nurnberg, G. K.: The prediction of internal phosphorus load in Kunz, M. J., Wüest, A., Wehrli, B., Landert, J., and Senn, D. B.: Im- lakes with anoxic hypolimnia phosphorus not considered in the pact of a large tropical reservoir on riverine transport of sediment, mass bal- sewage diversion Of six lakes, four showed model carbon, and nutrients to downstream wetlands, Water Resour. predicted, All the outliers had anoxic phorus concentration Res., 47, 1–16, https://doi.org/10.1029/2011WR010996, 2011. in anoxic Lake Sam- ma, Limnol. Oceanogr., 29, 111–124, Kunz, M. J., Senn, D. B., Wehrli, B., Mwelwa, E. M., and Wüest, https://doi.org/10.1016/j.scr.2014.10.005, 1984. A.: Optimizing turbine withdrawal from a tropical reservoir for Orlob, G. T.: Mathematical Modeling of Water Quality: Streams, improved water quality in downstream wetlands, Water Resour. Lakes and Reservoirs, John Wiley & Sons, Chichester, UK, Res., 49, 5570–5584, https://doi.org/10.1002/wrcr.20358, 2013. available from: http://pure.iiasa.ac.at/id/eprint/2144/ (last access: Langbein, W. and Durum, W.: The aeration capacity of 18 April 2019), 1983. streams, U.S. Dept. of the Interior, Geological Survey, O’Keeffe, J. H.: A Perspective on Training Methods Aimed at https://doi.org/10.3133/cir542, 1967. Building Local Capacity for the Assessment and Implementation Ledec, G. and Quintero, J. D.: Good dams and bad dams: of Environmental Flows in Rivers, Front. Environ. Sci., 6, 1–7, environmental criteria for site selection of hydroelectric https://doi.org/10.3389/fenvs.2018.00125, 2018. projects, Lat. Am. Caribb. Reg. Dev. Work. Pap. No. Olden, J. D. and Naiman, R. J.: Incorporating thermal regimes into 16, 21, available at: http://large.stanford.edu/courses/2011/ environmental flows assessments: Modifying dam operations to ph240/mina2/docs/Good_and_Bad_Dams_WP16.pdf (last ac- restore freshwater ecosystem integrity, Freshw. Biol., 55, 86– cess: 18 April 2019), 2003. 107, https://doi.org/10.1111/j.1365-2427.2009.02179.x, 2010. Padisák, J., Barbosa, F. A. R., Borbély, G., Borics, G., Chorus, I., Espindola, E. L. G., Heinze, R., Rocha, O., Törökné, A. K. and

www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019 1670 R. S. Winton et al.: Dams, water quality and tropical reservoir stratification

Vasas, G.: Phytoplankton composition, biodiversity and a pilot Fish. Soc., 105, 442–449, https://doi.org/10.1577/1548- survey of toxic cyanoprokaryotes in a large cascading reservoir 8659(1976)105<442:TOHSTV>2.0.CO;2, 1976. system (Tietê basin, Brazil), SIL Proceedings, 1922–2010, 27, Smith, V. H.: Eutrophication of freshwater and coastal marine 2734–2742, https://doi.org/10.1080/03680770.1998.11898164, ecosystems a global problem, Environ. Sci. Pollut. Res., 10, 126– 2000. 139, https://doi.org/10.1065/espr2002.12.142, 2003. Parker, F. L., Benedict, B. A., and Tsai, C.: Evaluation of math- Spoor, W. A.: Distribution of fingerling brook trout, Salvelinus ematical models for temperature prediction in deep reservoirs, fontinalis (Mitchill), in dissolved oxygen concentration gradi- National Envrionmental research center, US Environmental Pro- ents, J. Fish Biol., 36, 363–373, 1990. tection Agency, Corvallis, Oregon, 1975. Straskraba, M., Tundisi, J. G., and Duncan, A.: Comparative Petts, G. E.: Water quality characteristics of reg- reservoir limnology and water quality management, Springer- ulated rivers, Prog. Phys. Geogr., 10, 492–516, Science+Business Media, 1993. https://doi.org/10.1177/030913338601000402, 1986. Sutherland, A. B., Meyer, J. L., and Gardiner, E. P.: Effects of land Power, M. E., Dietrich, W. E., and Finlay, J. C.: Dams and Down- cover on sediment regime and fish assemblage structure in four stream Aquatic Biodiversity: Potential Food Web Consequences southern Appalachian streams, Freshw. Biol., 47, 1791–1805, of Hydrologic and Geomorphic Change, Environ. Manage., 20, https://doi.org/10.1046/j.1365-2427.2002.00927.x, 2002. 887–895, 1996. Tharme, R. E.: A global perspective on environmental flow assess- Preece, R. M. and Jones, H. A.: The effect of Keepit Dam on the ment: Emerging trends in the development and application of en- temperature regime of the Namoi River, Australia, River Res. vironmental flow methodologies for rivers, River Res. Appl., 19, Appl., 18, 397–414, https://doi.org/10.1002/rra.686, 2002. 397–441, https://doi.org/10.1002/rra.736, 2003. Premalatha, M., Abbasi, T., and Abbasi, S. A.: Science of the To- Thurston, R. V, Russo, R. C., and Phillips, G. R.: Acute tal Environment A critical view on the eco-friendliness of small Toxicity of Ammonia to Fathead Minnows, T. Am. hydroelectric installations, Sci. Total Environ., 481, 638–643, Fish. Soc., 112, 705–711, https://doi.org/10.1577/1548- https://doi.org/10.1016/j.scitotenv.2013.11.047, 2014. 8659(1983)112<705:ATOATF>2.0.CO;2, 1983. Price, R. E. and Meyer, E. B.: Water quality management Todd, C. R., Ryan, T., Nicol, S. J., and Bearlin, A. R.: The for reservoirs and tailwaters: operational and structural water impact of cold water releases on the critical period of post- quality techniques, Vicksburg, Mississippi, USA, available at: spawning survival and its implications for Murray cod (Mac- https://apps.dtic.mil/dtic/tr/fulltext/u2/a254184.pdf (last access: cullochella peelii peelii): A case study of the Mitta Mitta 18 April 2019), 1992. River, southeastern Australia, River Res. Appl., 21, 1035–1052, Richter, B. D.: Re-thinking environmental flows: from allocations https://doi.org/10.1002/rra.873, 2005. and reserves to sustainability boundaries, River Res. Appl., 30, Tolmazin, D.: Changing Coastal oceanography of the Black https://doi.org/10.1002/rra.1320, 2009. Sea. I: Northwestern Shelf, Prog. Oceanogr., 15, 217–276, Rivers-Moore, N. A., Dallas, H. F., and Morris, C.: To- https://doi.org/10.1016/0079-6611(85)90038-2, 1985. wards setting environmental water temperature guidelines: A Tundisi, J. G.: Distribuição espacial, seqência temporal e ciclo South African example, J. Environ. Manage., 128, 380–392, sazonal do fitoplâncton em represas: fatores limitantes e contro- https://doi.org/10.1016/j.jenvman.2013.04.059, 2013. ladores, Rev. Bras. Biol., 50, 937–955, 1990. Rolls, R. J., Growns, I. O., Khan, T. A., Wilson, G. G., Elli- Turner, R. E., Qureshi, N., Rabalais, N. N., Dortch, Q., Justic, D., son, T. L., Prior, A., and Waring, C. C.: Fish recruitment in Shaw, R. F., and Cope, J.: Fluctuating silicate:nitrate ratios and rivers with modified discharge depends on the interacting ef- coastal plankton food webs, P. Natl. Acad. Sci. USA, 95, 13048– fects of flow and thermal regimes, Freshw. Biol., 58, 1804–1819, 13051, https://doi.org/10.1073/pnas.95.22.13048, 1998. https://doi.org/10.1111/fwb.12169, 2013. UNEP: A Snapshot of the World’s Water Quality: Towards a Global Rubin, J. A., Gordon, C., and Amatekpor, J. K.: Causes and Assessment, available at: http://www.unep.org/publications/ (last consequences of mangrove deforestation in the Volta estuary, access: 18 April 2019), 2016. Ghana: Some recommendations for ecosystem rehabilitation, Van Cappellen, P. and Maavara, T.: Rivers in the An- Mar. Pollut. Bull., 37, 441–449, https://doi.org/10.1016/S0025- thropocene: Global scale modifications of riverine nutrient 326X(99)00073-9, 1999. fluxes by damming, Ecohydrol. Hydrobiol., 16, 106–111, Sato, Y., Bazzoli, N., Rizzo, E., Boschi, M. B., and Miranda, https://doi.org/10.1016/j.ecohyd.2016.04.001, 2016. M. O. T.: Influence of the Abaeté River on the reproductive Vannote, R. L. and Sweeney, B. W.: Geographic Analysis success of the neotropical migratory teleost Prochilodus argen- of Thermal Equilibria: A Conceptual Model for Evaluat- teus in the São Francisco River, downstream from the Três ing the Effect of Natural and Modified Thermal Regimes Marias Dam, southeastern Brazil, River Res. Appl., 21, 939–950, on Aquatic Insect Communities, Am. Nat., 115, 667–695, https://doi.org/10.1002/rra.859, 2005. https://doi.org/10.1086/283591, 1980. Selge, F. and Gunkel, G.: Water reservoirs: worldwide distribu- Walker, K. F., Hillman, T. J., and Williams, W. D.: Ef- tion, morphometric characteristics and thermal stratification pro- fects of impoundments on rivers: an Australian case cesses, in: Sustainable Management of Water and Land in Semi- study, SIL Proceedings, 1922–2010, 20, 1695–1701, arid Areas, edited by: Gunkel, G., Aleixo da Silva, J. A., and https://doi.org/10.1080/03680770.1977.11896755, 1978. Sobral, M. d. C., 15–27, 2013. Ward, J. V and Stanford, J. A.: Evolutionary Ecology of Aquatic Smith, L. L., Oseid, D. M., Kimball, G. L., and El-Kandelgy, Insects, Annu. Rev. Entomol., 97–117, 1982. S. M.: Toxicity of Hydrogen Sulfide to Various Life His- Weber, M., Rinke, K., Hipsey, M. R., and Boehrer, B.: tory Stages of Bluegill (Lepomis macrochirus), T. Am. Optimizing withdrawal from drinking water reser-

Biogeosciences, 16, 1657–1671, 2019 www.biogeosciences.net/16/1657/2019/ R. S. Winton et al.: Dams, water quality and tropical reservoir stratification 1671

voirs to reduce downstream temperature pollution and Young, W. C., Kent, D. H., and Whiteside, B. G.: The influence reservoir hypoxia, J. Environ. Manage., 197, 96–105, of a deep storage reservoir on the species diversity of benthic https://doi.org/10.1016/j.jenvman.2017.03.020, 2017. macroinvertebrate communities of the Guadalupe River, Texas J. Wera, F., Ling, T., Nyanti, L., Sim, S., and Grinang, J.: Effects of Sci., 27, 213–224, 1976. opened and closed spillway operations of a large tropical hydro- Zarfl, C., Lumsdon, A. E., Berlekamp, J., Tydecks, L., and Tockner, electric dam on the water quality of the downstream river, Hin- K.: A global boom in hydropower dam construction, Aquat. Sci., dawi J. Chem., 2019. 77, 161–170, https://doi.org/10.1007/s00027-014-0377-0, 2014. Wetzel, R. G.: Limnology: lake and river ecosystems, Academic Zhang, L., Li, Q.-H., Huang, G.-J., Ou, T., Li, Y., Wu, D., Zhou, Press, 2001. Q.-L., and Gao, T.-J.: Seasonal stratification and eutrophication Winemiller, K. O., McIntyre, P. B., Castello, L., Fluet-Chouinard, characteristics of a deep reservoir, Longtan Reservoir in subtrop- E., Giarrizzo, T., Nam, S., Baird, I. G., Darwall, W., Lujan, N. ical area of China, Huan Jing Ke Xue, 36, 438–47, available at: K., Harrison, I., Stiassny, M. L. J., Silvano, R. A. M., Fitzger- http://www.ncbi.nlm.nih.gov/pubmed/26031068 (last access: 19 ald, D. B., Pelicice, F. M., Agostinho, A. A., Gomes, L. C., October 2018), 2015. Albert, J. S., Baran, E., Petrere, M., Zarfl, C., Mulligan, M., Zhong, Y. and Power, G.: Environmental impacts of hydroelec- Sullivan, J. P., Arantes, C. C., Sousa, L. M., Koning, A. A., tric projects o n fish resources, Regulated Rivers: Research Hoeinghaus, D. J., Sabaj, M., Lundberg, J. G., Armbruster, J., & Management, 12, 81–98, https://doi.org/10.1002/(SICI)1099- Thieme, M. L., Petry, P., Zuanon, J., Vilara, G. T., Snoeks, J., 1646(199601)12:1<81::AID-RRR378>3.0.CO;2-9, 2015. Ou, C., Rainboth, W., Pavanelli, C. S., Akama, A., van Soes- bergen, A., and Saenz, L.: Balancing hydropower and biodiver- sity in the Amazon, Congo, and Mekong, Science, 351, 128–129, https://doi.org/10.1126/science.aac7082, 2016. Winton, R. S., Calamita, E., and Wehrli, B.: Physical data for the 54 most voluminous low latitude reservoirs, https://doi.org/10.3929/ethz-b-000310656, 2018.

www.biogeosciences.net/16/1657/2019/ Biogeosciences, 16, 1657–1671, 2019